Anthropic đã đấu thầu dự án 100 triệu USD về dàn đàn drone không người lái cho Pentagon, nhưng không được chọn, vị thế của SpaceX và OpenAI đã chiến thắng.

Theo Tin Tức 1M AI, Anthropic đã nộp một đề xuất earlier this year to participate in the $100 million Orchestrator Prize Challenge, a Pentagon project aiming to develop voice-controlled autonomous drone swarm technology. The competition, co-launched by the Defense Autonomous Warfare Group under Special Operations Command and the Defense Innovation Unit, is divided into five stages, progressing from software development to live-fire tests, with later phases focusing on "target perception and sharing" and "from launch to target." Anthropic's proposal, centered around Claude, translates commander intent into digital commands to coordinate drone swarms, excluding autonomous targeting or weapon decisions, with human oversight throughout, and the company also proposed a joint research program with the Pentagon to securely develop and assess autonomous weapon capabilities.

Anthropic believed its proposal did not cross its "no involvement in fully autonomous weapons" red line, as humans could still monitor and terminate the system at any time. However, Anthropic was ultimately not selected, with the reason unconfirmed by Bloomberg. Selected proposals included a joint submission by SpaceX and xAI, along with two defense tech companies that list OpenAI as an AI partner (one being a contractor for autonomous military vehicle testing, Applied Intuition). OpenAI's technology will be utilized in the "task control" segment to help convert voice commands into digital commands. Just hours after the Pentagon announced last Friday that its contractors were prohibited from engaging in commercial activities with Anthropic, OpenAI disclosed a new agreement with the Department of Defense to use its AI tools on a classified cloud system. Anthropic declined to comment. (Bloomberg)